gombru/multi-modal-hate-speech

Code of the paper "Exploring Hate Speech Detection in Multimodal Publications" accepted in WACV 2020.

Code of the publication "Exploring Hate Speech Detection in Multimodal Publications" accepted in WACV 2020.
Check the blog post for more information.
To be released soon.

Unfortunately I didn't have the time to clean and release this code. But given the interest shown, a messy version of the code can be found here. However, note that the interest of this research work is not the code and data modeling itself, but the dataset and the conclusions obtained in this very first approach to multi-modal hate speech detection. If you aim to reproduce the modeling explained in the paper, I suggest building on other and up to date classification projects which might address different tasks but present similar architectures. I wish the community a successful journey on this research!

multi-modal-hate-speech's Issues

Ids of original data split

Hi @gombru

Thank you for the dataset. Would you be able to make the splits folder available as well? It's not included in the current zip. I am actually running some experiments on your dataset and it would be useful if I have the split you used for reporting results in your paper for comparison and consistency purposes.

Thanks,
Gaurav.

Where is the Code?

Hi any update on the code?

Would be good to highlight the legal basis of the processing and republishing of personal data

As far as I can tell the data allow the association of individual twitter users with the data. Therefore the GDPR applies. It would be great also for third parties to understand the legal basis of processing particularly when others would try reusing any results within this repo.

I am currently involved in a case of a company who wants to make use of the data but probably e.g. would not fall under under article 89.